Frequent samples of blood were obtained for serum testosterone, LH, PRL, cortisol, and GH determinations in healthy men while awake and during nocturnal and daytime sleep. The results were correlated with polygraphic monitoring of the electroencephalogram, electrooculogram, and electromyogram. The mean serum levels of testosterone and PRL during nonrapid eye movement (NREM) sleep and rapid eye ... [Show full abstract] movement (REM) sleep differed significantly from those during the waking period. However, the mean levels of these hormones during REM sleep were not statistically different from those during NREM sleep. Through all states of waking and sleep, the mean level of LH remained unchanged. When the periods of nocturnal sleep were divided into four 3-h periods, namely 2000-2300 h (A), 2300-0200 h (B), 0200-0500 h (C), and 0500-0800 (D), testosterone was significantly increased during period D (P<0.05). LH levels were similar in all four periods. The PRL level was significantly elevated during periods C (P<0.01) and D (P<0.01). When the nocturnal waking state was also divided into the same four time periods, testosterone was significantly elevated in period D (P<0.05). In contrast, neither an increase in PRL nor any change in LH was seen during the nocturnal waking state. A marked elevation of cortisol was seen in period D (P<0.01) both during the nocturnal sleeping and the nocturnal waking states. During acute sleep-wake cycle reversal performed in two subjects, PRL was elevated during sleep in the daytime, but testosterone was not significantly altered. The GH level in these two subjects increased during the deep NREM stages during both nocturnal and daytime sleep. The present findings suggest that the rise in the serum level of testosterone during the early morning is not directly related to sleep or to any of the sleep stages or to the changes of LH or PRL, but may be derived from an endogenous circadian rhythm similar to that of cortisol.